We intend to produce and market a statistical software package for exact nonparametric inference on a broad range of problems commonly encountered by applied statisticians. The emphasis is on providing a computationally efficient, robust, and portable set of tools for performing exact inference whenever the large sample assumptions justifying conventional asymptotic inference are in doubt. This happens often with contigency tables that are characterized by empty cells or cells with small entries. Exact nethods for analyzing these "sparse" contingency tables, while preferred, cannot be applied because they have not been incorporated into any of the commercial statistical packages. This gap in commercially available software is not an oversight. Formidable technical difficulties are encountered when attempting to produce software for exact inference. Several major innovations in the theory of statistical computing have been published recently by the principal investigator and co- investigator. As a result, exact statistical inference is now a parctical proposition for most important types of contingency table data. It is our intention to integrate these innovations into a standalone statistical package for mainframe, mini, and microcomputers. The package will interface in a friendly way with popular commercial statistical packages like SAS, BMDP, and SPSS, and with LOTUS 1-2-3, a popular micro-computer spreadsheet.